XpIsUp

It seems that Craig Larman is pushing the XP is UP really line at ECOOP this year.

See http://www.ecoop2002.lcc.uma.es/T01.htm

In the abstract he claims that: "Contrary to some misperception, the UP encourages an agile approach" and further, that: "...most XP and Scrum practices and values are either part of the UP, or specializations of more general UP guidelines."

Is anyone here going to ECOOP? (SomeAnonymousPerson I presume)

Seriously folks, as RobertCecilMartin points out (http://www.objectmentor.com/resources/articles/RUPvsXP.pdf) RUP is so general you can disguise XP in a RUP wrapper. That's at the practices level obviously. (Could you disguise a RUP project in an XP wrapper? Now there's a challenge!)

As far as the values go, I'm not sure that RUP has any values (can anyone correct me?) but if there are any, they're hidden pretty well by comparison to XP. Indeed, if you compare the misinformational blurb on the Rational web site (e.g. http://www.rational.com/media/products/rup/tp183.pdf) - remember the UP is so erm, unified, because it's defined by one company - to any of the XP information in the public domain, it becomes apparent that:

• XP values individuals and interactions over processes and tools much more than RUP does
• XP values working code over comprehensive (modelling) documentation much more than RUP does

And (IMHO) that's what makes XP agile in a way that I believe that RUP can never be.

Uncle Bob's thesis puzzles me: the only circumstance under which I can see it being useful is if a team wants to do XP but must be seen to do RUP. That might come in useful some time, I guess. But really, what's the big deal?

And the claims for (R)UPs agile capabilties seem risible. Fine, so UP is iterative: make the iterations short. Is UP flexible enough to handle two-week iterations? So, UP is incremental: make the increments small. Is UP flexible enough to handle a few hours work as an increment? Out of the box UP orders incremental deliveries by "risk", whatever that is. Is UP flexible enough to handle increments ordered by value? Well, maybe it is. Whether or not any given organisation that's signed up to RUP would let a team do these things, that's something else.

Then the big one: is UP flexible enough to handle evolutionary deisgn? All the UP materials I've seen gloss over this in a big way. They explain about iterations (ie "lots of little waterfalls", they actually say this in places), and increments, then they point out of the window and say Hey, what's that?, and while everyone's looking away they slip in the e-word and continue as if evolutionary development were a necessary consequence of iterative, incremental working. Whereas it's almost the other way around: to be able to do evolutionary design (which you need to do in order to flatten the cost-of-change curve) you need to have many iterations and small increments, and you need easily changed, communally owned work products, and blah blah blah.

It looks a lot to me as tuning up UP to be as extreme as possible would be good, but that to do XP under the UP banner means, well, exactly that: do XP, but call it UP. (SomeAnonymousPerson again?!)

-- Doing XP but calling it RUP to keep your manager sweet (or your manager's manager) is the premise of Uncle Bob's paper. And that's the only reason I can see you why you would want to go down that route. TimBacon

Larman's tutorial at XP2002 was very interesting. Seems that those, like him, who take the UP seriously are really, really pissed off with all the Rational suites going around telling people evil nonsense.

The way he runs a full-on UP project, as described in the tute, is quite similar to what most people doing XP without an on-site customer (most of us) do. A couple more documents, perhaps. A few more diagrams with a longer lifetime that you might expect. Nothing too alarming.

This was quite an eye-opener for me after the horror stories I've seen under the UP banner. -- KeithB

Thinking about it some more, Larman's approach seem to be the opposite of Bob's. Bob's position seem to be that if you work in a UP shop you can get away with doing XP by using such-and-so arguments to show that XP is UP compliant. Larman's argument is that if you were doing UP right in the first place it would be sufficently like XP anyway that there wouldn't be a problem. --KB

Latest news: ObjectMentor have developed an XP "plugin" for Rational's RUP toolset, see http://www.rational.com/products/rup/xprog.jsp "Unlike books, articles or other passive reference materials, only the RUP Plug-In for XP provides specific, practical guidelines for XP development using the principles of RUP. " -- OliBye has to give them credit, this is one way of making money, if only somebody wanted to buy my soul.³host³³date³September 18, 2002³agent³Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.0; Q312461; H010818; T312461; .NET CLR 1.0.2914)³XpIsUp